Skip to content

Repository files navigation

🧬 Top Journal Paper Hub

中文版 · English


🇨🇳 中文版

简介

top-journal-paper-hub 是一套面向顶级期刊投稿的全流程 AI 写作 agent,覆盖从选刊、文献调研、论文起草、图表生成、引用核验、投稿前模拟审稿、投稿前检查,到审稿意见回复的完整链路。

支持 OpenClaw、Claude Code、Codex 三种平台,输出格式支持 LaTeX(Overleaf)和 Word(.docx),并内置 CSV/Excel 自动生图、CrossRef 实时引用核验、个人文献库 RAG 写作风格锚定。

v1.2.0 新增: Claude Code 插件打包、nature-paper-hub 单入口布局、nature-integrity-check、485 篇清洗文献索引,以及可选 LITREVIEW_API 私有文献库接入。


功能一览

Skill 功能说明
🏠 nature-paper-hub 主入口,12 阶段(0–11)全流程对话路由(含完整性检查与模拟审稿)
🔬 nature-reviewer-sim 投稿前模拟审稿:7 个专家 agent + 编辑决定信
🛡️ nature-integrity-check 投稿前/修回后完整性核验:引用、数据、图表、必备声明(pre-review / final-check)
📊 nature-figure CSV/Excel → 顶刊风格 matplotlib/R 图,自动识别图表类型
📖 nature-reader PDF/DOI/arXiv → 双语精读 Markdown,含图表解读与引用建议
📚 nature-citation CrossRef 实时核验 + RetractionWatch 撤稿检查 + BibTeX/RIS/ENW/Zotero 导出
🎞️ nature-paper2ppt 论文 → 中文/双语 PPTX,含演讲者备注

支持期刊(19 个)

Nature 系列(12 个)

期刊 影响因子 接收率 正文字数 图数 引用
Nature 63.7 ~8% 3,000 6 30
Nature Energy 60.9 ~8% 3,000 6 50
Nature Materials 37.2 ~9% 3,000 6 50
Nature Catalysis 37.8 ~8% 3,000 6 50
Nature Methods 32.1 ~8-10% 3,000 6 50
Nature Machine Intelligence 23.9 ~8% 3,500 6 50
Nature Synthesis 20.0 ~8% 3,000 6‡ 50
Nature Chemistry 19.2 ~9% 3,000 6 50
Nature Communications 15.7 ~20% 5,000† 10 60
Nature Chemical Engineering 13.0 ~8% 3,500 6+10ED 50
Nature Computational Science 12.0 ~12% 5,000 8 60
Nature Sustainability 25.1 ~10% 4,000 8 60

† Nature Communications 字数含 Methods 章节 ‡ Nature Synthesis 不接受 Scheme,只接受 Figure;Methods 中不可含图表

其他顶刊(7 个)

期刊 影响因子 接收率 文章类型 正文字数 图数
Science 44.7 ~7% Article / Report 4,500 / 2,500 6 / 4
Cell 45.5 ~8% Article / Short Article 8,000 / 4,000 8 / 4
PNAS 11.1 ~18% Research Article 6,000 6
JACS 14.4 ~11% Article / Letter ~6,000 / 4pp 8 / 4
ACS Nano 15.8 ~10% Article / Letter 10,000 / 4,000 10 / 5
Angewandte Chemie 16.1 ~10% Research Article / Communication 5,000 / 2,500 8 / 4
Advanced Materials 27.4 ~9% Research Article / Communication 5,000 / 2,500 8 / 4

安装

方式一:Claude Code 插件 marketplace(推荐,一键安装)

# 添加 marketplace(一次性)
/plugin marketplace add https://github.com/Yang1Bai/nature-paper-hub

# 安装插件
/plugin install nature-paper-hub

# 重新加载
/reload-plugins

安装后 7 个技能(主入口 + 6 个子技能)自动可用,无需手动 cp。


手动安装前的关键规则

不要只复制单个 SKILL.md。这些 skills 会读取仓库根目录的 scripts/templates/data/skills/nature-paper-hub/references/。手动安装时请保留完整 clone,并设置 CLAUDE_PLUGIN_ROOT 指向该 clone。

方式二:Codex(完整 skill 目录 + 根资源)

git clone https://github.com/Yang1Bai/nature-paper-hub.git
cd nature-paper-hub

mkdir -p ~/.codex/skills
cp -R skills/nature-* ~/.codex/skills/

export CLAUDE_PLUGIN_ROOT="$(pwd)"
pip install -r scripts/requirements.txt

Windows PowerShell:

git clone https://github.com/Yang1Bai/nature-paper-hub.git
cd nature-paper-hub

New-Item -ItemType Directory -Force "$env:USERPROFILE\.codex\skills" | Out-Null
Copy-Item -Recurse -Force .\skills\nature-* "$env:USERPROFILE\.codex\skills\"

$env:CLAUDE_PLUGIN_ROOT = (Get-Location).Path
py -3 -m pip install -r .\scripts\requirements.txt

方式三:Claude Code(手动 skills)

git clone https://github.com/Yang1Bai/nature-paper-hub.git
cd nature-paper-hub

mkdir -p ~/.claude/skills
cp -R skills/nature-* ~/.claude/skills/

export CLAUDE_PLUGIN_ROOT="$(pwd)"
pip install -r scripts/requirements.txt

方式四:OpenClaw

git clone https://github.com/Yang1Bai/nature-paper-hub.git
cp -R nature-paper-hub ~/.openclaw/workspace/skills/
# 重启 OpenClaw,对话中说"选刊"即可激活

验证安装

python -m unittest discover -s tests

手动安装后请不要删除 clone;CLAUDE_PLUGIN_ROOT 指向的目录就是脚本、模板和文献索引的资源根目录。


配置(可选)

本工具开箱即用,无需任何密钥。若你有自己的文献库 API,可设置环境变量 LITREVIEW_API(一个接受 ?q=<query> 的基础 URL)接入;不设置时,文献检索自动走「本地索引 + CrossRef + 网络搜索」,全流程不受影响。

export LITREVIEW_API="https://your-host.example.com/api/search"   # 可选

上手示例见 examples/quickstart.md


快速上手

OpenClaw / Telegram

在对话中输入以下关键词直接跳转对应阶段:

选刊 / choose journal  → 选择目标期刊(19个可选),显示字数/图数/引用限制
文献综述 / literature  → 调用个人文献库 + 网络搜索
写大纲 / outline       → 按期刊生成定制化论文结构
写摘要 / abstract      → 起草符合期刊风格的 Abstract
图表规划 / figure plan → 规划每张图的叙事逻辑
模拟审稿 / reviewer simulation → 投稿前3位审稿人视角模拟评审
检查引用 / citations   → CrossRef 实时核验 + 撤稿检查
导出 / export          → 选择 LaTeX(Overleaf)或 Word
写回复信 / rebuttal    → 逐条审稿意见生成回复框架

Claude Code

Use nature-paper-hub to start writing a paper for Nature Materials
Use nature-reviewer-sim to simulate peer review of my manuscript
Use nature-figure to generate a matplotlib figure from my data.csv
Use nature-reader to create a bilingual reader for this paper: ~/Downloads/paper.pdf
Use nature-citation to verify my reference list and export as BibTeX
Use nature-paper2ppt to convert this paper into a Chinese presentation

全流程说明

阶段 0:  选刊             → 19 个顶刊可选,加载对应字数/图数/引用限制
阶段 1:  文献调研         → 本地文献索引 +(可选)个人库 + CrossRef + 网络搜索
阶段 2:  结构规划         → 按期刊类型定制论文大纲
阶段 3:  逐节起草         → Abstract / Introduction / Results / Discussion / Methods
阶段 4:  图表规划         → CSV/Excel 自动出图 + matplotlib/R 代码 + 图例文字
阶段 5:  引用核验         → CrossRef API + RetractionWatch + 多格式导出
阶段 6:  投稿前完整性检查 → 引用/数据/图表/必备声明(pre-review 模式)
阶段 7:  模拟审稿         → 7 个专家 agent + 编辑决定信 + 修改路线图
阶段 8:  投稿前检查       → 字数/图数/格式/内容 checklist + cover letter
阶段 9:  最终完整性检查   → final-check 模式,零 CRITICAL 方可导出
阶段 10: 导出             → LaTeX (Overleaf) / Word (.docx) / PPTX
阶段 11: 审稿回复         → 逐条回复框架 + 修改说明 + 封面信

注:模拟审稿(阶段 7)可在任意阶段触发。

文件结构

nature-paper-hub/
├── README.md                              # 本文件(中英双语)
├── .claude-plugin/
│   ├── plugin.json                        # Claude Code 插件清单 🆕
│   └── marketplace.json                   # 一键安装 marketplace 清单 🆕
├── .github/
│   └── workflows/
│       └── journal-spec-check.yml         # 每周自动检查期刊规格 URL 健康状态
├── skills/
│   ├── nature-paper-hub/                   # 主 skill(hub,唯一入口)
│   │   ├── SKILL.md                       #   12 阶段路由骨架
│   │   └── references/                     #   按需加载的详细规则 🆕
│   │       ├── writing-rules.md            #     逐节写作规则 + 自评
│   │       ├── outlines.md                 #     论文大纲模板
│   │       ├── audit-and-cover-letter.md   #     投稿前 checklist + cover letter
│   │       ├── rebuttal.md                 #     审稿回复流程
│   │       ├── literature-integration.md   #     文献检索 / RAG / CrossRef
│   │       └── journal-special-rules.md    #     各刊特殊规则
│   ├── nature-reviewer-sim/SKILL.md       # 投稿前模拟审稿(7 agent + 编辑)
│   ├── nature-integrity-check/SKILL.md    # 投稿前/修回后完整性核验 🆕
│   ├── nature-figure/SKILL.md             # 科研绘图(自动识别 + matplotlib/R)
│   ├── nature-reader/SKILL.md             # 双语论文精读器
│   ├── nature-citation/SKILL.md           # CrossRef 引用核验 + 多格式导出
│   └── nature-paper2ppt/SKILL.md          # 论文转 PPT
├── templates/
│   ├── journal-specs.json                 # 19 个期刊规格数据
│   └── nature-latex.tex                   # Overleaf LaTeX 模板
├── data/
│   └── papers-index.json                  # 485 篇清洗后的文献索引
└── scripts/
    ├── auto_figure.py                     # CSV/Excel → 顶刊图
    ├── clean_papers_index.py              # 文献索引清洗脚本 🆕
    ├── export_docx.py                     # Word 导出
    ├── export_pptx.py                     # PPT 导出
    └── requirements.txt                   # Python 依赖

与同类项目对比(按最新版 main 校验)

对比基准:2026-06-21 检查 GitHub 默认分支;两个对比仓库均未发布 GitHub release,因此以当前 main 为准:Yuan1z0825/nature-skills 5d2ba1dee1c0,Boom5426/Nature-Paper-Skills 44cff42ac22a。若对方后续更新,应重新核对本表。

维度 top-journal-paper-hub Yuan1z0825/nature-skills Boom5426/Nature-Paper-Skills
期刊覆盖 19 个:Nature 系列 + Science / Cell / PNAS / ACS / Wiley Nature / Nature Communications + 通用 Nature-style skill;未见 19 刊索引 Nature 系列稿件栈;不追求全期刊覆盖
总控入口 nature-paper-hub 单入口,阶段 0-11 线性路由 多个 nature-* skill;README 未提供统一 hub paper-workflow 顶层路由
文献/RAG 底座 内置 485 篇清洗文献索引;可选 LITREVIEW_API 私有库 nature-academic-search + nature-citation;未见内置清洗索引 citation-verifier + reference-audit-guide;未见内置清洗索引
写作/润色 选刊、综述、起草、润色、投稿、返修一体化 nature-writingnature-polishingnature-response scientific-writingmanuscript-optimizerrebuttal-response
投稿前模拟审稿 nature-reviewer-sim(7 专家 agent) nature-reviewer paper-reviewer
投稿前完整性核验 nature-integrity-check 覆盖引用、数据、图表、必备声明 分散在 nature-datanature-citationnature-reviewer submission-audit + data-availability + citation-verifier
引用与撤稿检查 CrossRef 实时核验 + RetractionWatch/撤稿搜索 CrossRef / PubMed / arXiv / Scopus / ScienceDirect;未见 RetractionWatch 专项 CrossRef / Semantic Scholar 引用核验;未见 RetractionWatch 专项
图表工作流 CSV/Excel → 自动生图;matplotlib/R 投稿级图 nature-figure 支持 Python/R 投稿图;强调源数据可追溯 figure-planner 侧重图文结构和引用检查,未见自动生图脚本
可交付格式 LaTeX / Overleaf、Word .docx、PPTX PPTX、专利 DOCX、LaTeX 排版支持;未见通用论文 Word 导出 LaTeX/Overleaf 参考与演示支持;未见通用论文 Word 导出
Claude Code 打包 .claude-plugin/ 插件清单 + marketplace 元数据 可用于 Claude Code,但未见 .claude-plugin/ 清单 提供 Claude Code 安装说明,但未见 .claude-plugin/ 清单
Codex / OpenClaw Codex 兼容;OpenClaw / Telegram 兼容 Codex 优先;另含 OpenClaw medical skill 适配 Codex + Claude Code 兼容

更新日志

v1.2.0(2026-06)

  • 修复主 SKILL.md 阶段编号(重复 STAGE 7 / 错位 8.5)→ 统一为 0–11 线性流程
  • 打包为 Claude Code 插件(.claude-plugin/,支持 marketplace 一键安装)
  • 清洗 papers-index.json:534 → 485(去专利/碎片/重复,修复 OCR 乱码,补全期刊/年份)
  • frontmatter 同步 v1.2.0、description 覆盖全部 19 刊;README 补齐 nature-integrity-check
  • 修复脚本 bug:auto_figure.py box 分支不可达、export_docx.py 死代码

v1.1.0(2025-05)

  • 新增 7 个期刊:Science、Cell、PNAS、JACS、ACS Nano、Angewandte Chemie、Advanced Materials
  • 新增 nature-reviewer-sim 投稿前模拟审稿技能
  • 新增 GitHub Actions 每周期刊规格 URL 健康检查
  • 期刊总数:9 → 19

v1.0.0

  • 首发:12 个 Nature 系列期刊,5 个核心技能

用户反馈 💬

如果你用过这个工具写过论文,欢迎在 留言板 分享体验。


致谢

设计灵感部分来源于:


许可证

MIT License — 自由使用、修改和分发,保留原始署名即可。


贡献指南

欢迎提 Issue 和 PR!请说明:

  1. 目标期刊或功能
  2. 功能缺口或 bug 描述
  3. 建议实现方式


🇬🇧 English

Introduction

top-journal-paper-hub is a full-pipeline AI writing agent for top-journal submissions. It covers every stage from journal selection, literature review, manuscript drafting, figure generation, citation verification, pre-submission mock review, and reviewer response.

Compatible with OpenClaw, Claude Code, and Codex. Outputs LaTeX (Overleaf-ready) and Word (.docx). Features automatic figure generation from CSV/Excel, CrossRef real-time citation verification, and optional RAG-enhanced writing grounded in your own literature library (configurable via the LITREVIEW_API env var; the tool works fully without it).

v1.2.0: Added Claude Code plugin packaging, the single-entry nature-paper-hub layout, nature-integrity-check, a cleaned 485-paper index, and optional private-library integration via LITREVIEW_API.


Features

Skill Description
🏠 nature-paper-hub Main hub with 12-stage (0–11) conversational pipeline (incl. integrity check + reviewer sim)
🔬 nature-reviewer-sim Pre-submission mock review: 7 specialist agents + Editorial Decision Letter
🛡️ nature-integrity-check Pre-submission / post-revision integrity check: citations, data, figures, required statements (pre-review / final-check)
📊 nature-figure CSV/Excel → publication-quality matplotlib/R figures
📖 nature-reader PDF/DOI/arXiv → bilingual annotated Markdown with figure grounding
📚 nature-citation CrossRef verification + RetractionWatch + BibTeX/RIS/ENW/Zotero export
🎞️ nature-paper2ppt Paper → Chinese/bilingual PPTX with presenter notes

Supported Journals (19)

Nature Series (12)

Journal IF Accept Words Figs Refs
Nature 63.7 ~8% 3,000 6 30
Nature Energy 60.9 ~8% 3,000 6 50
Nature Materials 37.2 ~9% 3,000 6 50
Nature Catalysis 37.8 ~8% 3,000 6 50
Nature Methods 32.1 ~8-10% 3,000 6 50
Nature Machine Intelligence 23.9 ~8% 3,500 6 50
Nature Synthesis 20.0 ~8% 3,000 6‡ 50
Nature Chemistry 19.2 ~9% 3,000 6 50
Nature Communications 15.7 ~20% 5,000† 10 60
Nature Chemical Engineering 13.0 ~8% 3,500 6+10ED 50
Nature Computational Science 12.0 ~12% 5,000 8 60
Nature Sustainability 25.1 ~10% 4,000 8 60

† Nature Communications word count includes Methods ‡ Nature Synthesis: figures only (no Schemes); Methods cannot contain figures or tables

Other Top Journals (7)

Journal IF Accept Article Types Words Figs
Science 44.7 ~7% Article / Report 4,500 / 2,500 6 / 4
Cell 45.5 ~8% Article / Short Article 8,000 / 4,000 8 / 4
PNAS 11.1 ~18% Research Article 6,000 6
JACS 14.4 ~11% Article / Letter ~6,000 / 4pp 8 / 4
ACS Nano 15.8 ~10% Article / Letter 10,000 / 4,000 10 / 5
Angewandte Chemie 16.1 ~10% Research Article / Communication 5,000 / 2,500 8 / 4
Advanced Materials 27.4 ~9% Research Article / Communication 5,000 / 2,500 8 / 4

Installation

Option 1: Claude Code plugin marketplace (recommended, one-command)

# Add the marketplace (one-time)
/plugin marketplace add https://github.com/Yang1Bai/nature-paper-hub

# Install the plugin
/plugin install nature-paper-hub

# Reload to apply
/reload-plugins

All 7 skills (hub + 6 sub-skills) become available automatically — no manual copying.


Manual Install Rule

Do not copy only individual SKILL.md files. These skills read root-level scripts/, templates/, data/, and skills/nature-paper-hub/references/. For manual installs, keep the full clone and set CLAUDE_PLUGIN_ROOT to that clone.

Option 2: Codex (full skill directories + root resources)

git clone https://github.com/Yang1Bai/nature-paper-hub.git
cd nature-paper-hub

mkdir -p ~/.codex/skills
cp -R skills/nature-* ~/.codex/skills/

export CLAUDE_PLUGIN_ROOT="$(pwd)"
pip install -r scripts/requirements.txt

Windows PowerShell:

git clone https://github.com/Yang1Bai/nature-paper-hub.git
cd nature-paper-hub

New-Item -ItemType Directory -Force "$env:USERPROFILE\.codex\skills" | Out-Null
Copy-Item -Recurse -Force .\skills\nature-* "$env:USERPROFILE\.codex\skills\"

$env:CLAUDE_PLUGIN_ROOT = (Get-Location).Path
py -3 -m pip install -r .\scripts\requirements.txt

Option 3: Claude Code (manual skills)

git clone https://github.com/Yang1Bai/nature-paper-hub.git
cd nature-paper-hub

mkdir -p ~/.claude/skills
cp -R skills/nature-* ~/.claude/skills/

export CLAUDE_PLUGIN_ROOT="$(pwd)"
pip install -r scripts/requirements.txt

Option 4: OpenClaw

git clone https://github.com/Yang1Bai/nature-paper-hub.git
cp -R nature-paper-hub ~/.openclaw/workspace/skills/
# Restart OpenClaw. Say "选刊" or "choose journal" to activate.

Validate Install

python -m unittest discover -s tests

After manual installation, keep the clone in place; the CLAUDE_PLUGIN_ROOT directory is the resource root for scripts, templates, and the literature index.


Configuration (optional)

The tool works out of the box — no keys required. If you have your own literature API, point the LITREVIEW_API environment variable at a base URL that accepts ?q=<query>. When it is unset (the default), literature search falls back to the local index + CrossRef + web search, and the full pipeline still works.

export LITREVIEW_API="https://your-host.example.com/api/search"   # optional

See examples/quickstart.md for an end-to-end walkthrough.


Quick Start

OpenClaw / Telegram

选刊 / choose journal          → Select from 19 journals; see word/figure/ref limits
文献综述 / literature          → Personal library + CrossRef + web search
写大纲 / outline               → Journal-specific manuscript structure
写摘要 / abstract              → Draft journal-style Abstract
模拟审稿 / reviewer simulation → Pre-submission mock peer review (3 personas)
检查引用 / citations           → CrossRef verify + retraction check
导出 / export                  → LaTeX (Overleaf) or Word
写回复信 / rebuttal            → Point-by-point reviewer response framework

Claude Code

Use nature-paper-hub to start writing a paper for Advanced Materials
Use nature-reviewer-sim to simulate peer review of my manuscript
Use nature-figure to generate a matplotlib figure from my data.csv
Use nature-citation to verify my reference list and export as BibTeX

Full Pipeline

Stage 0:  Journal Selection      → Choose from 19 journals; load word/figure/ref limits
Stage 1:  Literature Review      → Personal library + CrossRef + web search
Stage 2:  Outline Planning       → Journal-tailored manuscript structure
Stage 3:  Section Drafting        → Abstract / Intro / Results / Discussion / Methods
Stage 4:  Figure Planning         → Auto-plot from CSV/Excel + matplotlib/R code + legend text
Stage 5:  Citation Check          → CrossRef API + RetractionWatch + multi-format export
Stage 6:  Pre-review Integrity     → Citations / data / figures / required statements (pre-review)
Stage 7:  Reviewer Simulation     → 7 specialist agents + Editorial Decision Letter + roadmap
Stage 8:  Pre-submission Audit     → Word / figure / format / content checklist + cover letter
Stage 9:  Final Integrity Check    → final-check mode; zero CRITICAL required to export
Stage 10: Export                   → LaTeX (Overleaf) / Word (.docx) / PPTX
Stage 11: Reviewer Response        → Point-by-point reply + change summary + cover letter

Note: Reviewer Simulation (Stage 7) can be triggered at any point.

Repository Structure

nature-paper-hub/
├── README.md                              # This file (bilingual CN/EN)
├── .claude-plugin/
│   ├── plugin.json                        # Claude Code plugin manifest 🆕
│   └── marketplace.json                   # One-command install marketplace manifest 🆕
├── .github/
│   └── workflows/
│       └── journal-spec-check.yml         # Weekly journal spec URL health check CI
├── skills/
│   ├── nature-paper-hub/                   # Hub skill (single entry point)
│   │   ├── SKILL.md                       #   12-stage routing skeleton
│   │   └── references/                     #   Detailed rules, loaded on demand 🆕
│   │       ├── writing-rules.md            #     Per-section writing rules + self-critique
│   │       ├── outlines.md                 #     Manuscript outline templates
│   │       ├── audit-and-cover-letter.md   #     Pre-submission checklist + cover letter
│   │       ├── rebuttal.md                 #     Reviewer rebuttal workflow
│   │       ├── literature-integration.md   #     Literature search / RAG / CrossRef
│   │       └── journal-special-rules.md    #     Journal-specific rules
│   ├── nature-reviewer-sim/SKILL.md       # Pre-submission mock review (7 agents + editor)
│   ├── nature-integrity-check/SKILL.md    # Pre-submission / post-revision integrity check 🆕
│   ├── nature-figure/SKILL.md             # Figure generation (auto-detect + matplotlib/R)
│   ├── nature-reader/SKILL.md             # Bilingual paper reader
│   ├── nature-citation/SKILL.md           # CrossRef citation verification + multi-format
│   └── nature-paper2ppt/SKILL.md          # Paper to presentation
├── templates/
│   ├── journal-specs.json                 # 19-journal specs database
│   └── nature-latex.tex                   # Overleaf-ready LaTeX template
├── data/
│   └── papers-index.json                  # 485 cleaned literature index entries
└── scripts/
    ├── auto_figure.py                     # CSV/Excel → publication figure
    ├── clean_papers_index.py              # Literature index cleaning script 🆕
    ├── export_docx.py                     # Word export
    ├── export_pptx.py                     # PPTX export
    └── requirements.txt                   # Python dependencies

Comparison (checked against latest main)

Baseline: checked GitHub default branches on 2026-06-21. Neither comparison repository has a GitHub release, so this table uses current main: Yuan1z0825/nature-skills 5d2ba1dee1c0 and Boom5426/Nature-Paper-Skills 44cff42ac22a. Re-check this table after those projects update.

Dimension top-journal-paper-hub Yuan1z0825/nature-skills Boom5426/Nature-Paper-Skills
Journal coverage 19 journals: Nature series + Science / Cell / PNAS / ACS / Wiley Nature / Nature Communications + generic Nature-style skills; no 19-journal index found Nature-series manuscript stack; not intended as broad all-journal coverage
Top-level routing Single nature-paper-hub entrypoint with stages 0-11 Multiple nature-* skills; no unified hub shown in README paper-workflow top-level router
Literature/RAG base Built-in cleaned 485-paper index; optional LITREVIEW_API private library nature-academic-search + nature-citation; no bundled cleaned index found citation-verifier + reference-audit-guide; no bundled cleaned index found
Writing/revision Journal choice, literature review, drafting, polishing, submission, and rebuttal in one pipeline nature-writing, nature-polishing, nature-response scientific-writing, manuscript-optimizer, rebuttal-response
Mock review nature-reviewer-sim with 7 specialist agents nature-reviewer paper-reviewer
Integrity preflight nature-integrity-check for citations, data, figures, and required statements Distributed across nature-data, nature-citation, and nature-reviewer submission-audit + data-availability + citation-verifier
Citation and retraction checks CrossRef verification + RetractionWatch/retraction search CrossRef / PubMed / arXiv / Scopus / ScienceDirect; no RetractionWatch-specific workflow found CrossRef / Semantic Scholar citation checks; no RetractionWatch-specific workflow found
Figure workflow CSV/Excel → auto figure; matplotlib/R publication figures nature-figure supports Python/R publication figures and traceable source data figure-planner focuses on figure story and reference checks; no auto-figure script found
Deliverable formats LaTeX / Overleaf, Word .docx, PPTX PPTX, patent DOCX, LaTeX layout support; no general manuscript Word export found LaTeX/Overleaf references and presentation support; no general manuscript Word export found
Claude Code packaging .claude-plugin/ manifest + marketplace metadata Usable from Claude Code, but no .claude-plugin/ manifest found Claude Code install docs, but no .claude-plugin/ manifest found
Codex / OpenClaw Codex-compatible; OpenClaw / Telegram-compatible Codex-first; also includes an OpenClaw medical skill adaptation Codex + Claude Code compatible

Changelog

v1.2.0 (2026-06)

  • Fixed main SKILL.md stage numbering (duplicate STAGE 7 / misordered 8.5) → linear 0–11 pipeline
  • Packaged as a Claude Code plugin (.claude-plugin/, one-command marketplace install)
  • Cleaned papers-index.json: 534 → 485 (removed patents/fragments/dupes, fixed OCR author noise, backfilled journal/year)
  • Synced frontmatter to v1.2.0, broadened description to all 19 journals; README now lists nature-integrity-check
  • Fixed script bugs: unreachable box branch in auto_figure.py, dead code in export_docx.py

v1.1.0 (2025-05)

  • Added 7 journals: Science, Cell, PNAS, JACS, ACS Nano, Angewandte Chemie, Advanced Materials
  • New skill: nature-reviewer-sim — pre-submission 3-persona mock peer review
  • GitHub Actions: weekly journal spec URL health check with auto-issue on failure
  • Journal count: 9 → 19

v1.0.0

  • Initial release: 12 Nature-series journals, 5 core skills

User Feedback 💬

If you have used this tool in your research, please share your experience on the feedback board.


Acknowledgements

Inspired in part by:


License

MIT License — free to use, modify, and distribute with attribution.


Contributing

Issues and PRs welcome. Please include:

  1. Target journal or feature
  2. Feature gap or bug description
  3. Suggested implementation approach

About

Full-pipeline AI agent for Nature-series journal writing. Supports OpenClaw, Claude Code, and Codex. Covers journal selection, drafting, figure generation, citation verification, and reviewer response.

Topics

Resources

Contributing

Stars

17 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages